Agents previous response ID failure

API
,
1

I am trying to reuse responses between agents runs, like in

result = await Runner.run(agent, "What is the largest country in South America?")
    print(result.final_output)
    # Brazil

result = await Runner.run(
        agent,
        "What is the capital of that country?",
        previous_response_id=result.last_response_id,
    )

found in documentation

However, I get this error when running the pipeline trying to reuse the last response reasoning

Error getting response: Error code: 400 - {'error': {'message': 'Duplicate item found with id rs_<ID>. Remove duplicate items from your input and try again.', 'type': 'invalid_request_error', 'param': 'input', 'code': None}}. (request_id: <RID>)

Any clues to what I am doing wrong?

1 Like
2

Works fine for me. Make sure you have the latest updates by running:

`pip install --upgrade openai openai-agents`

You can also add that and other specific parameters like this:

agent = Agent(name="Assistant",
              model='gpt-5-mini',
              model_settings = ModelSettings(
                  reasoning={"effort":"minimal"},
                  extra_body={"previous_response_id":"resp_12345"}
                  ),
              )
3 Likes
3

Hmm, are you sure? You are setting a fixed response ID in the Agent initializer, while I am trying to use the response from the previous agent run.

The run method has a previous_response_id parameter, string type.

1 Like
4

I tried your code too, no errors. That agent code was an extra alternative. Have you tried updating the packages?

2 Likes
5

Ok thanks for testing. Yes I have upgraded packages. Weird thing is that it is the same duplicate rs_ prefixed ID supposedly being duplicate, while the response IDs are prefixed resp_ it seems. I might have hit a snag.

6

Here, I tried this way and it worked too:

from agents import RunConfig
agent = Agent(name="Assistant previous_id",
              model='gpt-5-mini',
              model_settings = ModelSettings(reasoning={"effort":"minimal"},),
              )
result = await Runner.run(agent, "What is the largest country in latam?",)
print(result.final_output)
result2 = await Runner.run(agent, "What is the capital of that country?",
                        run_config=RunConfig(model_settings = ModelSettings(
                        extra_body={"previous_response_id":result.last_response_id}
                  ),
              ))
print(result2.final_output)

If by area, the largest country in Latin America is Brazil (about 8.5 million km²). If you meant by population, it's also Brazil (about 215 million people as of the mid-2020s).

The capital of Brazil is BrasĂ­lia.

ps: check if the agent doesn’t have conflicting parameters

1 Like
7

Thanks, I would never have guessed the latest parameter example :slight_smile:

Still getting the error though, need to investigate more. BTW the agent that I want to feed the previous response is a handoff from the previous agent. That could mess things up perhaps.

8

You mentioned getting a “rs_xxxx” id, it seems strange.

Does this also give the same id?

result.raw_responses[-1].response_id
2 Likes
9

Yes, same ID. I suspect the problems come from the agent being a handoff of the previous, if not I can live without it. They do work together nicely, without passing previous response explicitly.

1 Like